Book Search Engine
A python project for finding books through their description.
Summary
A search enginge for finding books through their description. The user can use any word and provided there is a match, results will be shown with regard to the three vectrozation methodologies utilized. The results indicate that the last technique (deep learning model) is superior to the others since books and their descriptions are matched even when the query's words are not included in any book description.The three vectorization methodologies are:
- TF-IDF
- Elasticsearch's BM25
- Universal Sentence Encoder from Google
Original Data were acquired from:
Personal Contribution:
- Collected data from Kaggle, preprocessed book text descriptions in an appropriate format and loaded Universal Sentence Encoder from Tensorflow-hub to acquire semantically similar books.
- Led a team of three students by splitting each task according to each member’s strengths.
- Integrated the aforementioned models with ElasticSearch and constructed a Flask application with a simple UI to effectively communicate the project's value to non-technical stakeholders.
Team members:
Code is available at:
Written with:
Python, Tensorflow, Keras, Flask, Pandas, Scikit-Learn, Numpy, ElasticSearch